REVIEW OF STRIP AND BLOCK ADJUSTMENT DURING 1964-1967
7
errors. A detailed description of these
methods can be found in ref. [38] and [39].
Most of the authors use a method that be
longs to the first group. Especially simple is
the method of block adjustment given by
Roelofs [47]. Here, an internal block adjust
ment is performed in which scale, azimuth,
and shifts of the sections are adjusted sep
arately. Thompson [49] and Van der Weele
[50] describe the use of base lines in this ad
justment. Thompson’s paper is of additional
interest because it exposes the often read fal
lacy that some errors in strip triangulation
are by their nature of the third degree in the
^-coordinate.
References [38] and [39] describe a method
of the second group in which the coordinate
connections between models are enforced by
choosing the transformed coordinates of the
connecting points (or, rather, corrections to
their approximate values) as parameters.
This reduces the number of parameters from
seven to just over four per model.
References [41], [42], [52], and [53], too,
enforce the coordinate connection but they
use the well-known double summation of the
effect of transfer errors. Although this reduces
the number of parameters to those of one
model of a strip, it produces condition equa
tions for each control point and for each
tie point between strips in which corrections
to the transfers of scale, azimuth, and tilts
occur as corrections to quasi-observations.
Jerie, in the latter two references, has re
duced the complications which this causes in
the formation of the normal equations by the
use of smoothing procedures and fictitious
points.
Especially in the case of sparse ground con
trol, a provision for the elimination of system
atic errors in the strip triangulation should
be included. In [38], [52], and [53] this is
achieved by including second-degree terms in
the transformation. If one wishes to avoid this
contamination with the idea of polynomial
adjustment, the procedure in ref. [39] can be
followed. Here, the conditions that the trans
fer errors should be equal to zero are replaced
by the conditions that, at least in the case of
equal model widths, the transfer errors at
each two successive connections should be
the same.
Strip Triangulation
1. STRIP FORMATION FROM INDEPENDENT
MODELS
At several centres, the triangulation of
independent models is followed by strip
formation and polynomial strip- and block-
adjustment. Ref. [55] to [64] treat the strip
formation for that purpose.
The strip formation consists in connecting
each model to the preceding one by means of
a similarity transformation. In most cases,
an exact coordinate connection is made at the
common perspective centre. Very simple
formulas for this purpose are given by Thom
son [59], [60], and Schut [57].
Reference [62] gives the standard procedure
for determining the model coordinates of the
perspective centres from grid measurements
made at two heights. Ref. [55] describes the
computation of these coordinates by resec
tion, using measurements made at one height.
The latter computation requires pre-calibra
tion of the projection cameras.
Inghilleri and Gaietto [55] perform only
an approximate relative orientation in the
analog instrument. The adjustment of the
relative orientation, based upon recorded
parallaxes, is included in the strip formation.
2. TRIPLETS IN STRIP TRIANGULATION
References [65] to [68] describe two
methods of analytical strip triangulation
based upon the orientation of triplets. Ander
son and McNair perform independent orien
tations of the triplets. The triplets are joined
into strips by making the orientation of the
centre photograph of a triplet and the bx of
its first model equal to those obtained for
this photograph and for this base component
in the preceding triplet. Keller and Tewinkel
perform the triplet orientation while enforcing
the orientation of its first photograph and
the strip coordinates of the points whose
images lie across the centre of this photo
graph.
Consequently, with both methods the man
ner in which two successive triplets are con
nected and as a result the strip deformation
caused by errors and by deformation of the
photographs depends upon the direction of
triangulation. This can be avoided by follow
ing McNair’s recommendation [66] to connect
successive independent triplets by similarity
transformations using all common points.
It has been claimed that triplet triangula
tion results in a stronger or more rigid strip
than triangulation by independent relative
orientation and scaling of successive models.
However, Moellman [69], using C&GS pro
grams, reports that after second- or third-
degree polynomial strip adjustment there is
no way to distinguish between the results of
the two triangulations. McNair [66] has ob
tained better results from his triplet triangu-